What is search term pattern grouping?

Search term pattern grouping is the practice of reading the search terms report and sorting wasted queries into buckets by the reason they are irrelevant — wrong intent, wrong product, wrong audience, wrong stage — then writing one negative that blocks the whole bucket instead of one negative per query. For example, dozens of queries containing free, cheap, or diy usually share a single intent problem: the searcher will not pay. One phrase negative for each of those words handles all of them, current and future, rather than negating each specific phrase as it appears.

How is pattern grouping different from n-gram analysis?

N-gram analysis is token-frequency math: you count how often each word or two-word phrase appears across your search terms and their cost, and the high-cost tokens surface as negative candidates. Pattern grouping is the intent step that comes after — you look at why a cluster of queries is junk and decide the match type and scope of the negative. N-grams tell you which words appear a lot; pattern grouping tells you whether free means block everything or just a specific offer, and whether to use a phrase or exact negative. They are complementary, not competing.

What match type should a pattern negative use?

Most pattern negatives should be negative phrase, because a pattern is defined by a word or ordered phrase appearing anywhere in the query. A negative phrase for free blocks every query containing that word in that order — free trial, download for free, free shipping — which is exactly the point of a pattern. Reserve negative exact for cases where only one specific query is junk and its close variants are fine. Avoid negative broad for patterns unless you fully understand its reach, because it can block queries that do not contain your term at all.

Won't a pattern negative block converting traffic?

It can if you group by surface wording instead of by intent, which is the main risk. Before you commit a pattern negative, filter the search terms report to every query containing that token and confirm none of them convert. If free appears in both junk queries and a converting free consultation term, the pattern is too coarse — tighten it to a two-word phrase like free download that isolates the junk. Grouping saves time only when the group is genuinely all waste; a lazy group throttles good traffic the same way any careless negative does.

How often should I run pattern grouping?

Run a quick pass weekly on your top-spending campaigns and a deeper pass monthly. Because close-variant matching keeps widening what your keywords can reach, new queries that fit an existing pattern appear constantly — but if you have already negated the pattern, they never cost you anything, which is the whole advantage over single-term negation. The weekly pass is mostly confirming existing patterns still hold and catching new intent clusters; the monthly pass is where you review whether any pattern has grown too broad or too narrow.

Group search terms into patterns, not one-off negatives

Negating search terms one query at a time is a treadmill you never step off. You open the search terms report, spot free crm download, add it as a negative, and next week free crm tool, crm for free, and download crm free show up having cost you the same wasted clicks. Each is a slightly different string, so an exact negative on the first one never caught them. The problem was never those specific queries — it was the pattern they all share: someone searching with the word free is telling you they will not pay. Block the pattern once and every present and future member of it is handled.

Pattern grouping is the discipline of reading the report for intent clusters instead of individual strings. It is the step that turns the report from a list you react to into a set of rules you maintain. As one practitioner writeup of the workflow puts it, the goal is to stop treating the report as a stream of one-off decisions and start seeing the “patterns” that let you negate in bulk. This post is about how to find those clusters, write the one negative that kills each, and avoid grouping so coarsely that you block real demand.

Why single-term negation never catches up

Adding one negative per junk query fails for a structural reason: the supply of unique wasteful queries is effectively infinite, while the number of intents behind them is small. A searcher who wants something free can phrase that in hundreds of ways, but they are all the same intent. When you negate free crm download as an exact negative, you have blocked exactly one of those hundreds. The other phrasings keep matching your broad and phrase keywords, keep spending, and keep forcing you back into the report to negate the next variant. You are always one step behind the queries because you are fighting strings, not intents.

Close-variant matching makes this worse every quarter. Google expands what a keyword can match to include synonyms and reformulations it judges similar, so the surface of queries your keywords reach keeps widening even when you change nothing. That is the same mechanism that hides new wasted spend the top-cost view misses: fresh junk queries arrive continuously. Single-term negation treats each arrival as a new task. Pattern negation treats the whole future stream of one intent as already handled, which is the only way the maintenance cost stops growing with your account.

Grouping by the reason a query is junk

The unit of pattern grouping is the reason a query wastes money, not the words it contains. Read down your search terms report sorted by cost and, for each query with spend and no conversions, ask which bucket it belongs to. Most junk falls into a handful of intent buckets that repeat across almost every account:

Won’t-pay intent — free, cheap, diy, template, open source. The searcher is looking for a way to avoid buying.
Wrong product — queries for an adjacent thing you do not sell. If you sell project management software, project management jobs and project management certification are whole categories, not one-offs.
Research / job-seeker intent — salary, jobs, course, tutorial, what is, meaning. These are people who will not become customers this session.
Competitor or brand mismatch — a cluster of queries naming a rival or a brand you do not stock, which usually deserves its own decision rather than a blanket block.

Once a query is in a bucket, you are no longer asking “should I negate this string” but “what single term defines this bucket.” That reframing is what makes the work finite. A report with two hundred wasted queries usually collapses into eight or ten intent buckets, and eight or ten phrase negatives dispatch the lot. The buckets are also stable: the won’t-pay bucket looks the same in a plumber’s account and a SaaS account, so once you have built the habit the grouping goes fast.

Writing one negative that kills the bucket

Having named a bucket by its defining word, you write a single negative — almost always negative phrase — on that word. A negative phrase for free blocks every query where those letters appear as a word in order, so free trial, crm free, and get it free are all caught by one entry. This is the entire payoff: the pattern negative matches queries you have never seen yet, because it keys on the intent marker rather than the full string. You have converted an endless list of reactions into one rule.

Match type is where pattern negatives go wrong, so choose it deliberately. Use negative phrase for genuine patterns, because a pattern is exactly “this word, anywhere.” Reserve negative exact for the rare case where one specific query is junk but its close variants convert. Avoid negative broad for patterns unless you have thought through its reach, since a negative broad keyword can suppress queries that do not even contain your word. The mechanics of how each negative match type decides what it blocks — including the plurals and close-variant gotchas that catch people out — are covered in negative keyword match types, and they apply to pattern negatives exactly as they do to any other.

Where n-gram analysis fits

Pattern grouping and n-gram analysis are two halves of the same job, and the strongest workflow uses both. N-gram analysis is the mechanical surfacing step: you break every search term into its constituent words and short phrases, sum the cost and conversions behind each token, and let the high-cost, zero-conversion tokens rise to the top. It is how you discover that free has quietly cost you a few hundred dollars across forty different queries you would never have connected by eye. The method and how to build it are in n-gram search term analysis.

Pattern grouping is the judgement step that follows. An n-gram table tells you free is expensive; it does not tell you whether to block it. That decision needs a human to check whether free ever appears in a converting query, whether the right block is the single word or a two-word phrase like free download, and which match type to use. Run the n-gram pass to find the candidate tokens, then group and adjudicate them into pattern negatives. The n-gram gives you the shortlist; pattern grouping turns the shortlist into safe, durable rules.

The over-grouping trap

The failure mode of pattern grouping is grouping by surface wording instead of true intent, which turns a time-saver into a traffic-killer. The word free is a reliable won’t-pay marker in most accounts, but free consultation might be one of your best converting queries. Block free as a phrase negative without checking and you have silenced a money-making term along with the junk. A pattern is only safe when every query in it is genuinely waste, so the discipline is to verify the group before you commit the negative.

The verification is quick and non-negotiable: before adding a pattern negative, filter the search terms report to every query containing that token and scan the conversion column. If the token is pure junk, negate the single word. If a converting query hides in there, tighten the pattern to a two-word phrase that isolates the waste, or drop to a narrower set of negatives. Over-grouping is the same mistake as over-negating in general, where negatives quietly suppress demand you wanted — the broader consequences of that are in when negative keywords cost you sales. Grouping multiplies your leverage, which means it also multiplies the damage of a careless group.

Making it a weekly habit

Pattern grouping is worth building into a fixed cadence because its value compounds. Run a short weekly pass on your highest-spend campaigns: sort the search terms report by cost, confirm your existing patterns are still holding, and look for any new intent cluster that has emerged. Most weeks you will find that the queries which slipped through fit a bucket you already have, which means the system is working — those cost you nothing because the pattern was already blocked. The weekly job shrinks over time instead of growing, which is the opposite of what happens with single-term negation.

Do a deeper monthly pass to maintain the patterns themselves. Over a month, check whether any pattern has drifted — a word that was pure junk when you blocked it may now appear in a new converting query as your offering changes, or a bucket you split too finely could be consolidated. This is also where the pattern library connects to the rest of your account structure: the negatives you build belong in a deliberate hierarchy of account, campaign, and shared lists, covered in how to structure negative keyword lists, and the search terms report you are reading is the same one described in the guide to the search terms report. Grouping is the method; the report is the input and the list structure is where the output lives.